My research is focused on statistical analysis of protein structure, development of methods for structure prediction of proteins and protein complexes, and applications of computational structural biology to problems in cancer research. At the most basic level, we apply modern methods of statistical analysis to protein structural parameters such as bond angles and dihedral angles for use in structure determination and structure prediction software. We also seek to understand the physics behind the distributions and interdependencies we observe in experimental structures. I run a Molecular Modeling Facility, enabling my colleagues in experimental cellular and molecular biology to develop and test hypotheses in biological systems with a knowledge of the three-dimensional structures of the proteins they work on. From the available data for any target of interest, we aim to develop the most biologically informative models possible, whether that comprises an oligomer and/or complex with small ligands or nucleic acids, or different conformational and functional states. To do this well, we need to understand the biological assemblies, both homo- and heterooligomers, observed in protein crystals. We utilize a structural bioinformatics approach, seeking evidence for particular interactions and assemblies in multiple crystal forms across a family (or family pair) of proteins. These studies are also relevant to predicting and interpreting the functional relevance of missense mutations identified in gene sequencing in clinical settings. Because of their relevance to cancer therapy, we are studying all of the available structures of antibodies and kinases (the two most common domains in the Protein Data Bank) in order to understand their structural and functional variation. In the case of kinases, we are interested in the conformational changes involved in trans autophosphorylation, and how inhibitors might be developed to interfere with this process, especially for mutated kinases. In the case of antibodies, we are developing methods for computational antibody design to both globular proteins and disordered or denatured proteins with linear epitopes, both as therapeutics and reagents for molecular biology studies.
Cancer is responsible for 23% of all deaths in the United States in recent years. Gene sequencing of patient tumors and peripheral blood is providing unprecedented levels of information on molecular drivers of cancer. Protein structural information can be used to understand the mechanism of cancer driver mutations and to develop therapeutics aimed at common mutations in some genes.
